0
$\begingroup$

Suppose $\{a_n\}$ is a sequence such that $\lim_{n\to\infty} (a_{n+1}-a_n)=0$.

Prove that if there exists $c>0$ such that for all $n$, $|a_n|\geq c$ then either for all but a finite number of terms $a_n>0$ or for all but a finite number of terms $a_n<0$.

I fail to see how $\lim_{n\to\infty} (a_{n+1}-a_n)=0$ helps here. I know that it means that for all $\varepsilon>0$, eventually $|a_{n+1}-a_n|<\varepsilon$. I tried playing with $\varepsilon$ by setting it to $c$ and use the triangle inequality but with no luck. Any tips?

$\endgroup$
1
$\begingroup$

HINT: If there are infinitely many "changes of signs" (i.e., if the conclusion doesn't hold), you can construct a subsequence of $\{a_{n+1}-a_n\}_n$ which doesn't converges to zero. Use the condition $|a_n|\geq c$.

The best way you can construct this sequence is this: Suppose $a_1$ is negative, let $a_{n_1+1}$ be the FIRST positive number in the sequence after $a_{1}$ (it exists by assumption), then let $a_{n_2+1}$ be the FIRST positive number in the sequence after $a_{n_1+1}$ and so on... argue that $a_{n_k}$ is negative for all $k$, $\{a_{n_k}\}$ will be your sequence. (A similar argument works if $a_1$ is positive).

$\endgroup$
  • $\begingroup$ I tried proving by contradiction. Can you elaborate please? What subsequence do you mean? Thanks for your reply. $\endgroup$ – user228929 Apr 5 '15 at 16:03
  • $\begingroup$ I claim that there exists a subsequence $\{a_{n_k}\}_k$ of $\{a_n\}$ such that $a_{n_k+1}a_{n_k}<0$ (i.e. $a_{n_k+1}$ and $a_{n_k}$ have opposite signs). Can you argue this? Once you have this, I'm sure you can prove that $|a_{n_k+1}-a_{n_k}|>2c$ for all $k$. (HINT: Suppose $a_{n_k+1}$ is possitive and $a_{n_k}$ is negative, then $a_{n_k+1}-a_{n_k}$ is a positive number so it's equal to its absolute value and $|a_{n_k+1}|=a_{n_k+1}$, $-a_{n_k}=|a_{n_k}|$). $\endgroup$ – Daniel Apr 5 '15 at 16:19
  • $\begingroup$ Let me verify: suppose the statement is false. Then there exist infinitely many positive elements of the sequence and infinitely many negative ones. So if $a_n>0$ for some $n$, I can find $N>n$ such that $a_N<0$. By taking such pairs of positive-negatives we get a subsequence $\{b_n\}$ of $\{a_n\}$. Then if we suppose $b_{n+1}>0$ and $b_n<0$, $|b_{n+1}-b_{n}|=|b_{n+1}|+|b_{n}|\geq 2c$. That's more or less what I did. But the initial condition with limit applies to $\{a_n\}$ and not $\{b_n\}$. Where am I wrong? $\endgroup$ – user228929 Apr 5 '15 at 16:59
  • $\begingroup$ You're right until this point: "If $a_n>0$, I can find $N>n$ such that $a_N<0$. I don't know if it's clear but there is must be an $i$ between $n$ and $N$ (inclusive) such that $a_i$ is positive and $a_{i+1}$ is negative (there must be a change of sign somewhere). Your final argumentis fine. $\endgroup$ – Daniel Apr 5 '15 at 17:05
  • $\begingroup$ I edited the answer. Just in case. $\endgroup$ – Daniel Apr 5 '15 at 17:10

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.